Review the recommendation. Record the human decision.
Teranode reviews a recommendation, whether drafted by an AI or a person, using models from separate vendors. It flags what the firm's designated reviewer should examine and produces a timestamped artifact with the reviewer's human-entered attestation.
Self-review is not a separate control.
A model can repeat its own assumptions. A team can repeat the assumptions behind its own recommendation. Teranode identifies the source, excludes its model vendor when one exists, and assigns a separate review panel.
Four lenses across at least three model vendors.
Evidence, fiduciary and conflicts, risk and best-interest fit, and alternatives lenses return attributed, flag-form findings. Strict schemas, fail-closed completion, and deterministic aggregation keep the model boundary explicit. No single AI makes the final call.
Every finding receives a human disposition.
The reviewer marks action required, records that the answer addressed the finding, disagrees with the finding, or marks it not applicable. Any outcome other than Action required requires a reviewer-written reason. For Action required, the system records a standard concurrence on selection. The final decision remains human and must be consistent with the recorded dispositions.
A timestamped artifact with human-entered attestation.
The record preserves source and reviewer provenance, findings, limitations, dispositions, reviewer-supplied name and role, final decision, timestamp, Teranode archive status, and a content-comparison fingerprint. Teranode does not verify reviewer professional status. The SHA-256 fingerprint is a comparison value, not a digital signature.
Finance first. Design-partner slots open.
Teranode Review is currently designed for RIA compliance workflows. The next product evidence is an accountable firm workflow and an identified reviewer, not an unsupervised public decision tool.